Data warehouses are not easy to build. Their design requires a way of thinking at odds with the manner in which traditional computer systems are developed. Their construction requires radical restructuring of vast amounts of data, often of dubious or inconsistent quality, drawn from numerous heterogeneous sources. Their implementation strains the limits of today's information technology. Not surprisingly, a very large number of data warehouse projects fail.
We meet people in a wide variety of industries who want to build data warehouses. Given the challenges they will face, we start by making sure the effort is worth the trouble. One of the first questions we always ask is, “Why do you want to build a data warehouse?” Frequently, we get replies like the following:
“Because we need to clean up our customer data.”
“To centralize enterprise data in a single repository.”
“So that we can write queries that come back from the database quickly.”
“To keep our historic data on line.”
Each of these answers focuses on a technological component or process. While some of them may be among the valid technical objectives of a data warehouse, none of these answers are reason enough to build one. Unless there are better reasons that have gone unstated, each of these projects will likely fail. Technology for its own sake serves no one.
Successful data warehouses are built for one reason: to answer business questions. The nature of the questions to be addressed will vary, but the intention ...